You are on page 1of 13


On the prehistory of the Arabic language

Federico Corriente Crdoba Universidad de Zaragoza
Dealing with the pre-Islamic period of Arabic is nearly the same as making guesses at what the
prehistory of a language could have been like, on the mere basis of a few statements issued by
neighbouring people who did not speak it and of a host of scattered epigraphic materials which, however,
may or not reflect the speakers true speech and, at any rate, yield very little trustworthy information
because of the well-known shorthand features of most Semitic scripts, compounded in this case by their
being encoded in makeshift adopted Aramaic or South Arabian unvocalized alphabets.
This much said, it is only fair to commend Dr. Mascitelli1 for his enthusiastic resolve to undertake an
almost impossible task, while fully conscious of this, in an area where many a seasoned Semitic scholar
has failed or, at least, been unable to produce new conclusive evidences. We would wholeheartedly admit
our feeling of having often sailed in that same boat and regularly failed in the same purpose in many, if
not perhaps all of our pronouncements on this matter; however, we also think that those of us who have
joined this fray do not deserve the epithet of fools for having rushed into grounds where smarter and more
angelical fellows have refrained from treading. Because, Arabic being in many ways the best known and,
therefore, the most important Semitic language, any increase in our present degree of knowledge thereof is
of paramount importance for a better description of the whole family; as a matter of fact, the very scarcity
of certainties about that pre-Islamic and prehistoric period of its existence weighs heavily on the whole
realm of Semitic linguistics, and has often been the source of bad mistakes and misapprehensions. The
slightest shade of a chance of success in this endeavour is in our view, therefore, worth the effort and the
risk of failure, as meant by the Arabic proverb /ill aiyyah fal aliyyah/ if (the purpose) is not attained,
let it (at least) not be untried. In such disposition only are we trying to review this new important addition
to the bibliography on the oldest phases of Arabic; at any rate, the momentous impact of whatever
opinions are published on an issue like this is the main justification for the unusual length of our notes and
This published version of the authors Ph.D. dissertation is divided in two distinctly outlined parts,
namely, a first one, devoted to the emergence, background and spreading of Arabic in the pre-Islamic
period, and a second one, being a selection of basic epigraphic texts, followed by their graphemic,
phonetic and morphological analysis and a final chapter on the origins of the Arabic script.

1. D. Mascitelli, Larabo in epoca preislamica. Formazione di una lingua (Arabia antica 4), Roma 2006, LErma di
Bretschneider, 14 x 21, 337 pp.

Aula Orientalis 25 (2007) 141-153



Both main parts, 1 and 2, are characterized by an exhaustive survey of former opinions, with no other
conspicuous absence than that of some first-rate Russian authors and works, such as Anna Belovas
Istorieskaja morfologija arabskogo jazyka po materialam pamjatnikov doislamskogo perioda
(Historical morphology of the Arabic language on the basis of its documents in the pre-Islamic period,
Moscow, Vostonaja Literatura, 1994),2 Oerki po istorii arabskogo jazyka (Sketches of history of the
Arabic language, same place and editor, 1999) and Ximjaritskij jazyk (Himyaritic language, same
place and editor, 1996),3 and D. Frolovs Classical Arabic Verse. History and theory of ar (Leiden
Boston Kln, 2000);4 however, in this matter the author sins in very good company, or perhaps should
we say in the company of the majority, limited as we all are in our capacity for at least reading every
language in the world with an important scientific production. Otherwise, the bibliography listed and
frequently used is quite complete and updated, which is remarkable and commendable in an epoch when
students and even professors not only shun the Classical languages, but allow themselves to ignore
German, Italian and Spanish, even French in some countries. The Medieval Western European scholars
saying Graece, non legitur has turned for them into non legitur nisi Anglice, with the sad
consequences that could be expected from such ignorance and narrow-mindedness.
It is also noteworthy that our colleague mentions the most relevant scholars opinions on every issue,
but rather seldom adopts or rejects them, except in a few cases which we shall comment on as they come
up. This attitude is most coherent in the case of the second part of his book, where the edition,
commentary and interpretation of the sample texts either produce the impression of a definitive solution or
of an avowed puzzling conundrum, but we feel that the statements contained in the first part would call
more often for a larger measure of involvement and criticism although, of course, the author is entitled to
let the reader draw his own conclusions and take any of the available options. Thus, e.g., Retss view of
the term Arab (pp. 32-33), as the label of a peculiar community or brotherhood of initiated fellows
giving military protection to farmers and tradesmen, may well be fascinating, but clashes openly with the
historical and even anthropological evidence for its being an ethnic designation since its earliest
appearances and it is unlikely to provide a clue for distinguishing true ethnic Arabs from mere Arabic
In other instances, contrariwise, we would say that Mascitelli has paid too much heed to certain
school tenets which he ought perhaps to have considered with some degree of scepticism. Such would be
the cases, in our view:
1) of the classification of Old North Arabian epigraphic material into two groups (pp. 34, fn. 11, and
42), according to the shape of the definite article, >h-< or >h(n)-<, respectively, it being obvious that the
optionally extended shape is a mere phonetic variant, probably explainable in diachronic and/or diatopic
terms, but insufficient by itself to posit very different dialects or linguistic phases. The case might be quite
another when that shape is >l/m-< (p. 225), which necessarily betrays a less akin dialect without,
however, thereby precluding mutual intelligibility.5
2) of the rejection of linguistic connotations for the traditional classification of Arabs into tribes of
Qanian and Adnnian descent (p. 44 and, most emphatically, in p. 146, where it is described as a
2. Reviewed by us in Estudios de Dialectologa Norteafricana y Andalus 2(1997)242-243.
3. Reviewed by us in Estudios de Dialectologa Norteafricana y Andalus 2(1997)244-245 and 4(2000)238-241, respectively.
4. Reviewed by us together with its Russian original (Klassieskij arabskij stix, Moscow, Nauka, 1991) in Journal of Arabic
Literature 31.3 (2000)267-273. Frolovs data and views might have helped Mascitelli in improving his outline in p. 63, fn. 39,
upon discussing the eventual places of origin of verse in pre-Islamic Arabia.
5. Cf. the case of the Balearic dialect of Catalan, where the shape of the definite article is s(a) (< Latin ipse - ipsa), instead of
el la (< Latin ille illa), which does not hinder understanding between heterogeneous speakers, as continues to be the case also
in Modern Yemenite dialects with a/im/l- shapes for that functional.



political, rather than a scientific and philological operation), following on this view no lesser a scholar
than Goldziher. Our colleague declares the acceptation of those connotations as risky, if not useless,
though right away pronouncing the tribal attribution of linguistic facts and features as relevant; however,
as we said in a survey of South Arabian features in Andalusi Arabic one gets the impression, after
carefully studying the linguistic peculiarities attributed to the various tribes by native authors, that there
was indeed a certain correlation between their dialects and origins since, as a matter of fact, South Arabian
features do appear in the speech of tribes considered to be Qanian and viceversa.6 To give just one
example for the sake of concision, would it be wise to disconnect the consistently reported attribution of
the relative pronoun //, an exact match of South Arabian >(w)<, to the tribe of ayyi from their
reported Qanian extraction? One must, of course, contemplate the distinct possibility of some South
Arabian individuals and even communities having on occasions entirely forsaken their former language
and thoroughly mastered one of the North Arabian dialects and the literary koine; however, the survival of
some South Arabian grammatical and lexical items in their speech is equally or even more expectable, as
given away by instances like those mentioned in that article of ours.7 As a matter of fact, our most recent
studies on the emergence of Western Arabic unequivocally point to Egypt and the process of
decreolization of the important Yemenite settlement there.8
3) of his assumption of the classification of Arabic as Central Semitic (pp. 18-19 and 51), an
original contribution of Hetzrons in line with Garbinis innovative views on the subgrouping of Semitic.
In our view, however, neither of both hypotheses are more than fads, not standing serious verification, as
we have demonstrated in two recent papers.9
4) of his reluctance to admit true diglossia in pre-Islamic Arabia (p. 57) after having,
nevertheless, correctly assumed the existence of dialects then and there, as well that of a literary koine.
Under such circumstances, the only possible conclusion is that a significant number of people, not of
course the whole population, were diglottic, i.e., had a more or less good command of that koine, in
addition to one of their native dialects, no matter whether these were urban or rural, Bedouin or else. We
must assume that not everybody in pre-Islamic Arabia could probably understand and enjoy a qadah,
on account of register differences, imagery and allusions, but this should not be purported as proof of
cultural diversity, it being simply, in the eyes of the contemporary society, a case of sheer lack of culture.
Exactly the judgment which most of us would pass on many an Englishman of our days who would not
fully understand Shakespeares idiom, it being obvious, furthermore, than in any diglottic community
there is a direct relation between higher degree of culture and truer diglossia and vice versa.
5) of his suggestion of a trichotomy in the linguistic situation of pre-Islamic Arabia, whereby there
would have been a spoken language, a written language and a literary language. The fact that epigraphic
materials, although by their own nature always written, do not often conform themselves to the
requirements of the literary oral koine simply reflects their authors lack of competence or interest in using
this highest level, if indeed they had reached it chronologically or geographically. As in the case of Middle
Arabic, those performances did not belong to any of the well-defined systems which are usually labelled
6. In our paper South Arabian features in Andalusi Arabic, in Studia lingistica et orientalia memoriae Haim Blanc
dedicata, Wiesbaden, Harrassowitz, 1989, 94-103, esp. 94, fn.1.
7. The lexical issue was dealt with as early as by C. Rabin 1984, On the probability of South Arabian influence on the Arabic
vocabulary, in Jerusalem studies on Arabic and Islam 4, 125-134.
8. See our paper On the degree of kinship between Andalusi Arabic and Maltese..., in Folia Orientalia 41 (2005) 25-38,
esp. p. 36, fn. 34. We shall come back again to this theory in a forthcoming more detailed paper.
9. Namely, On the degree of kinship between Arabic and Northwest Semitic, in AIDA 5th Conference proceedings, Cadix,
2002, 187-194 and Lexicostatistics and the Central Semitic theory, in apal tibnim m illak. Studies presented to Joaqun
Sanmartn on the occasion of is 65th birthday. Aula Orientalia-Supplementa 22 (2006) 139-144.



as language, and were just cases of hesitation or even intended choice between the spoken dialects and the
literary koine. On the other hand, one should beware of generalizations, such as that of Mascitelli in p. 20,
when he states that only very recently have Arabic dialects been used for written and literary purposes,
thus forgetting the Andalusi zaal, profane and mystical, and proverb collections of the Middle Ages.10
6) of his disposition to accept a revised theory of noun declension as a perhaps adventitious addition
to Arabic morphology (p. 68-70). Professor Owens proposals on this issue are, of course, well-reasoned
on several particular accounts, but the general coincidence of, at least, Akkadian, Ugaritic, Arabic and
Ethiopic in the shape and functions of most case morphemes should be more than sufficient to posit noun
declension as a common Proto-Semitic feature and refrain from toying with more original and modern but
less convincing and well-grounded hypotheses. There is no real proof to say, e.g., (p. 72) that pre-Islamic
Arabic possessed noun declension, but this was most likely used in an irregular way, even not at all in the
urban speech. Let us not forget that Vollers hypothesis never won the day, as the basic coincidences
between the language of pre-Islamic and Proto-Islamic poetry and the Qurnic text are largely sufficient
to prove than noun declension followed the Classical rules most regularly in the high registers of prose
and verse, while the trust placed by early grammarians in some Bedouins competence in such matters11
can only be construed as an attestation of the regular presence of that feature in the middle registers of
some Post-Islamic communities, at least until the 10th century, according to Ibn inn.12 Besides, even
some of the texts included in Mascitellis part 2 (chapter I, pp. 105 and 110) contain evidences of irb
(i.e., noun declension) used in total agreement with Classical Arabic rules, such as >wl ns1yhm< = /walinisihim/ and for their women, and >wl yhmw< = /wali-ihim/ and for their
brothers:13 we cannot understand why then the author says about this latter text (p. 111) that its
attribution to Arabic is very dubious.
7) Our colleagues faith in other renowned scholars infallibility is again probably excessive when
he gives credit (p. 65) to Monroes description of the language of Modern Bedouin poetry as a nearclassical koine, understood by illiterate, and descending directly from the ancient poetical koine, three
qualifications which call for some important structures, namely, that the koine of the so-called naba
poetry is uninflected Neo-Arabic, wholly understandable only to people used to its peculiar idiom, and not
so directly derived from the pre-Islamic and Proto-Islamic poetical koine that it had not considerably
renewed its imagery and lexicon. Neither would we agree with our colleague, and with the most
knowledgeable dialectologist W. Diem, whom he is following on this matter, when they consider that in
the dialectal pairs /bintak bintik/ your daughter, /abk abki/ your father with gender distinction in
the possessor, final short inflectional vowels have survived phonetically, although devoid of their old
syntactic functions: in the second pair, the feminine suffix is actually a surviving Proto-Semitic {-k},14
designed to avoid gender confusion, which would otherwise have happened even in Old Arabic in pause,
while in the first pair we are rather confronted with a case of insertion by infixation of a former suffix. The
solution is not different from that of /bintu(h)/ his daughter and /binta(h)/ her / his daughter in the
same or other dialects, which must not be derived from Old Arabic /bintuh - bintah/, through
10. A comprehensibe bibliography about both subjects can be found in Corriente 1997, Poesa dialectal rabe y romance en
Alandals, Madrid, Gredos, pp. 374-383, with some important additions in more recent years.
11. According to Blau 1963, deservedly mentioned in Mascitellis bibliography. This positioning of his on this issue reappears
in p. 246, invoking the support of Shahid 1980, which again cannot alter the well-established facts.
12. See our paper From Old Arabic to Classical Arabic, also mentioned in the authors bibliography, p. 66, fn. 3.
13. The broken pl. // is rather uncommon in Arabic, but is closely matched by Ethiopic /aw/, which in a text like
this, with a strong admixture of Sabaic, and coupled with the suffix >hmw< might point in that same direction.
14. See Brockelmanns Grundri I 309 about its original anceps quantity, and Murtonen 1964, An Introduction to the
comparative grammar of the Semitic languages, p. 109, about the occasional survival of similar instances in Hebrew, while the
length of that vowel is standard in Ethiopic.



defunctionalization of the nominative and accusative cases respectively, but again from a process of
infixation of the morphematic final vowels /u/ und /a/, by the same morphological phenomenon which
generated internal plurals and feminines in South Semitic15 and, more recently, the infixed 3rd person pl.
masc. of the perfective (i.e., /ktawb/ they wrote) and 2nd person sg. fem. of the imperfective (e.g.,
/tr kz/, vs. the masc. /tr
kz/ you stand upright) in Modern South Arabian.16
However, when considering that the core of this survey is found in part 2, where the author expounds
the core of his dissertation, i.e., the selected texts, which are the skeleton and foundation of his
reconstruction of the oldest phases of North Arabian, we must say, in spite of the modest outcome
expectable and resulting from his praiseworthy endeavours, that his choice of samples is skillful and fair,
as it accommodates different layers of epigraphic evidences (in South Arabian, Nabataean and Arabic
scripts). They are sufficiently representative without superfluous duplication and illustrative of what these
materials can contribute to an improved and updated description of earliest Epigraphic North Arabian,
avowedly one of the most elusive topics in Semitic Linguistics on account not only of the sketchy and
iterative nature of the related mostly brief inscriptions, but also of an unyielding graphemic code and
constant interferences with Aramaic and South Arabian scripts and languages.
We shall not question the authors classification of the sources for the study of Old Arabic into direct,
which would be, according to him, only the epigraphic ones, and indirect, i.e., those recorded in Islamic
times, as this is part of the method chosen by him, although this choice forces upon him continuous and
not always founded doubts about the true Arabic character of certain words and constructions. We
understand that a long-standing relation with a given language tends to make us believe that we have
acquired the instinct of detecting what is genuine and what is alien to it, but does not confer us any
infallibility in such judgments, because we must always ignore wide areas of the past epochs or remote
recesses of that language which we believe to master, so that we may at any time be fooled by substratum,
adstratum and superstratum interferences as well as by semantic developments which we could not have
dreamt of.
Therefore, we are going to offer our reactions to his survey of sample texts as prospective
contributions to a better comprehension of them, without any pretence of superseding previous
interpretations. This includes some further areas of disagreement between us and Mascitelli, which we feel
obliged to air for the sake and benefit of contrasting opinions:
1) In some passages, we observe that a final {-h} would provide a better reading if interpreted as a
reflex of the 3rd person fem. morpheme of the perfective, more akin to Hebrew that to its Arabic match {at}, e.g., in >qrbh< she approached (p. 96), perhaps >h< (p. 104, if it is to be read as >uat<
she has been protected, and >bnh< (p. 117), best understood as /banat/ she built, with a fem. subject.
In such instances, our colleague would had benefited from taking into account Fleischs reports17 about
exchanges of final /t/ and /h/ in the dialect of ayyi, perhaps cases of pseudo-corrections betraying the
process of acquisition of North Arabian speech by this genealogically South Arabian tribe.
2) Some features of his transcription system, like: a) >a< and >at< for pausal and contextual t
marbah, instead of >ah< and >at<, which has become a common practice among many Western
Arabic scholars, but causes confusion in cases like the pausal forms of /ayh/ life and /ay/

15. See our Introduccin a la gramtica comparada del semtico meridional (Madrid, Consejo Superior de Investigaciones
Cientficas, 1996, p. 36 and fn. 3, and p. 39 and fn. 1.
16. See T.M. Johnstone, The Modern South Arabian Languages (Undena, Malibu, 1975), pp. 16-17.
17. In his Trait de philologie arabe I (Beyrouth, Imprimerie Catholique 1961, p. 185, also valid for cases where >lh< may
be more reasonably understood as the goddess Allt.



shyness,18 b) the generalization of >< for both >< and ><, which does away with the graphic
difference between many pairs of semantically diverse words and c), above all, from the same origin, the
habit of transcribing whole paragraphs with pausal forms, which is outright shocking when it happens in
the Qurn (p. 62, fn. 33, p. 83, p. 84, fn. 82, p. 179, etc.); such a practice can be excused, for the sake of
brevity in book titles and personal names, even in short technical expressions, but never in whole quotes
of sustained texts. One is equally surprised by our colleagues derivation of the personal names Sbawayhi
and Miskawayh from *Sibo and *Masko , instead of attested Pahlavi s b b
y and muk b
y, as
well as by his looking askance at transcriptions of Greek chi with Arabic n and viceversa, it being
notorious that the pronunciation of that Greek phoneme and, incidentally, its Russian counterpart, in front
of palatal vowels becomes very close if not identical with the German ich-Laut (//) and has, therefore,
been assimilated to Arabic //,19 c) some carelessness about the accurate transcription of several Arabic
items, like the name of the tribe Mai (always with /d/ in p. 103, 155, 157, 159, etc.), / ula /
Tuesday (p. 97, *al- al a), /nufas/ or /nafs/ woman in childbed (p. 99, *nufs), etc.
3) Some statements about the grammar of Arabic and its dialects may have been issued hurriedly,
without checking their accuracy. E.g., in p. 198, it is said that the phoneme // would be characteristic of
Classical Arabic, as against the dialects but, in fact, some of them kept it apart from //, without operating
the habitual merger, as shown in the older stages of Andalusi Arabic and in some modern dialects like that
of Da nah.20 In the same manner, Mascitellis statements about the pronunciation of t marbah in
pause as just /a/ (p. 74, 190 and 211-212) are inaccurate as far as Classical Arabic is concerned, as proven
by rhymes, where the resulting /-ah/ is matched by any other /h/, whether part of the root morpheme or
final phoneme of some pronouns. The phonemic reality of that /h/ was precisely the main reason for the
invention of the grapheme called t marbah and this is a synchronic fact which should not be obscured
by the complex diachronic merger of two different feminine markers, {t} and {}, or by a different
situation in younger phases of Arabic, where that /h/ has been in fact, at least phonetically, if not
phonemically, dropped. The phonotactic rules of Old and Modern Arabic do not tolerate a short vowel
abutting upon a final juncture, which caused their lengthening (e.g., in loose rhymes, or in the case of
the accusative marker an in pause) or the addition of /h/ or // (even in the case of long vowels at the end
of broken plurals of biconsonantal roots, e.g., /miyh/ waters, /im/ female slaves). Neither can we
agree with our colleagues assumption of a broken plural pattern *{afulah} (p. 99), or with his
proposal (p. 242, fn. 161) to explain the idle alif attached to the verbal plural morpheme {} as a device
to guarantee its lengthening, since it reappears not only optionally when that same morpheme is attached
to a noun, but also in the case of the accusative marker {an}, whether in pause or context, and is likelier
to be just a left-over from the vertical stroke separating words in South Arabian script, put to a new use as
a diacritic device to avoid confusion of homographs. Equally strange are Mascitellis Arabic reading
*/man azzun/ (p. 108), where, in the context of >mn *zzm wwnym< whether powerful or weak,
18. A curious feature of that procedure is that pausal forms are inserted everywhere in context, except when the final vowels
are not inflectional marks; in other words, its aim is to avoid the pitfalls of inflection at the expense of producing a distorted image
of any kind of Arabic, classical or dialectal. But this shortcut is not foolproof, as can be seen in p. 122, where min qabla is
ungrammatical, as there is only the adverb min qablu beforehand and the compound preposition min qabli before, but qabla is
impossible after min.
19. About which, see Grundri I 122. It is a pity that some Semitic scholars of our time, preferring fads to time-honoured
indisputable knowledge, do not give its full value to Brockelmanns and his contemporaries works, which remain useful and
correct in most of their statements and have not been systematically outdated by more recent and fashionable authors and views.
20. See our paper -L doublets in Classical Arabic as evidence of the process of de-lateralisation of d and development of
its standard reflex, in Journal of Semitic Studies 23.1(978)50-56, esp. p. 50, fn. 4 and 5, with references to our A grammatical
sketch of the Spanish-Arabic dialect bundle (Madrid, Instituto Hispano-rabe de Cultura, 1977), and J. Cantineaus Cours de
phontique arabe (Paris 1960, p. 56).



i.e., /min azzin wawnin/, his vocalizations *ar and * ar, in p. 168, for /ariya/ to be naked
and / ariya/ to covet, as well as his analysis of >bny wd < in the text 4 (p. 109-110), where he
seems not to recognize the Classical Arabic dual verbs /banay waada / they both built and
erected, spelled as it could only be expected in the Epigraphic North Arabian orthography, where /w/ and
/y/ do appear often, not only as graphemes of the matching diphthongs, but even as matres lectionis for //
and //, unlike the case of alif for //, which was still being inconsistently introduced much later in the
Qurnic orthography. Why, then, insisting on reading *alarab kullihi in the famous an-Namrah
inscription, which is reputedly very close to Classical Arabic, and look for complicated explanations of the
wrong congruence, when script and Old Arabic grammar would simply demand /alarabi kullih/ of all
the Arabs?
4) Generally speaking, Mascitelli does not pay sufficient attention to dialectal variegation in North
Arabian, i.e, Old Arabic. Thus, for instance, although the causative verbs had usually the prefix //, there
are some residual cases where /h/ is still used, and the same applies to Qurnic variants of the V
measure21 which, in p. 100, might have provided a better reading for >hr< as haarraa, i.e.,
Classical Arabic taarraa to implore. In the same manner, a deeper acquaintance with certain
intricacies of Arabic phonemics would have kept him from suggesting a VII measure verb from a root
beginning with hamz, like his proposed *inadana to be authorized, in p. 115;22 in the realm of
morphology, the same applies to sinn, which he labels as dialectal in that same p., while in fact the
oblique case of /sa/unna/, not *sinna, is /sinna/ in correct Classical Arabic, and to the repeated mistake
all, masc. sg., i.e., /alla/ (p. 230 and 233), and allt as feminine sg. of the relative pronoun (p. 230),
i.e., /allat/: that shape belongs in fact to the feminine pl.
5) Neither has been our colleague totally felicitous in his choice of the technical labels of linguistic
groups and periods, such as North Arabian, which he restricts to what used to be known as Epigraphic
North Arabian (p. 18), only to call this latter Old Arabic a bit later (p. 21), although showing his
preference for Proto-Arabic, quite reasonably this time; needless to say, this causes some confusion to the
reader confronted with an unusual semantic opposition between Arabic and North Arabian, e.g., in p. 101
and 103. The same applies sometimes to reshaped grammatical terms such as *nn al-muakkid in p. 101
(for nn al-takd), or alif hamza (p. 213), for alifu l-hamz or alifu / hamzatu l-qa, it being better not
to innovate Arabic grammatical terms, as often done by some Westerners, who have coined the weirdlooking and cacophonous fata-tanwn, kasra-tanwn, etc.23
6) At times we feel that the received interpretations of certain passages of the inscriptions can be
improved in the light of Standard Arabic usage; that would be the case, e.g., of >dky tmr s1my dm
wlr s2r<, where >dm< is clearly the collective of /dmah/ persistent rain, while >s2r< would not
be exactly barley, but the wild grass which grows in the desert when it rains, the general meaning being
as long as heavens pour continuous rains and the earth grows grass.24 This would also be the case of the
term /ub/, again in the inscription of an-Namrah (p. 157 and 159), about which our colleague is first
hesitant, and then takes the wrong course, in spite of his been aware of its meaning of sedentary
settlements in South Arabian, not nomadic tribes, possibly because he has overlooked the classical and
21. See W. Wright, A grammar of the Arabic language, I 36 and 38.
22. Arabic dictionaries contain only one such item, /inaar/ to be curved, as an allomorph of /taaar/.
23. About which, see our paper Las etimologas rabes en la obra de Joan Coromines, in Lobra de Joan Coromines,
Sabadell 1999, 67-87, esp. 73.
24. In fact, Classical Arabic has /air/ growing scrubs and /ar/ scrubland, so commonly used that it entered
Castilian as jara. This idiom belong to a well-known semantic structure in Classical Arabic whereby eternity is expressed by the
endless alternation of certain natural phenomena like sun and moon rises, night and dawn, rain and draught, etc. Even the famous
zajjl Ibn Quzmn used similar phrases, e.g., in 87/33 as long as light and darkness alternate, stars set and the moon rises.



endemic dichotomy of Qurn XLIX,13 / aalnkum uban waqabila/ we made of you

(sedentary) peoples and (nomadic) tribes.
7) A number of misprints, as is almost unavoidable, have crept into an otherwise beautifully edited
volume, of which we have detected *umm al-qarya and *al-qayra in p. 46 and fn. 25, for /umm al-qur/,
Baumstamme in p. 51, for Baumstamm, Corrente in p. 57, for Corriente, la iza (sic, perhaps an
untimely reminiscence of /izah/) in p. 84, fn. 81, for /iz/, Mbhhl in p. 110, for Mbhl, *za spear,
in p. 156, for /zu/, mairq in p. 205, fn. 54, for /mariq/, festehender, in p. 239, for feststehender.
8) The attribution of the technical term midianisch Thamudisch to us in p. 55 calls for an
explanation, as it is indeed in our paper of 1975, but only within a quote from Littmann.25
This long list of additions and corrections might give the reader of this review the impression that
Maschitellis work is weak or premature. This is not our view and, in fact, the number of our remarks is
not too high, when considering that this book contains more than three hundred pages on an extremely
problematic issue, as we have said repeatedly. We do commend his hard work in patiently harvesting the
materials, and handling them with appropriate methods, and we congratulate him for his ability to produce
in most points an accurate picture of what Pre-Islamic Old Arabic was, and how it developed into PostIslamic Old Arabic, Classical Arabic and Neo-Arabic.
The fact that P. Larcher begins the introduction of his manual26 by stating that this work is basically a
course on grammatical questions related to the Arabic verb must not be lost from sight upon commenting
on his sources, methods, attitudes and goals. As such a course, it is addressed to students in the main aim
of clarifying Classical usage, therefore assuming a predominantly synchronic and preceptive style, which
precludes any deep or extensive excursus in such areas as old and modern dialectology and, widely
speaking, diachrony. Those of us who have authored grammars of Arabic for our college students are
indeed familiar with that predicament, and can have only sympathy for other colleagues equally compelled
to engage in that pedagogical endeavour for practical reasons, such as the scarcity of works well-adapted
to students of Arabic with linguistic interests beyond mere competence and performance.
However, once a course is edited and made available to wider circles of users, it is assumed and
desirable that it be reviewed, which means being noticed by the scholarly community and receiving
support on its positive aspects and criticism in those felt to be improvable. This is what we shall try to do
in the next lines, at the risk of sounding nit-picking, cantankerous, old-fashioned or one-sided on several
issues, but only our frank opinion can mean some contribution to the author or readers in general, and that
is what we shall give them all.
To begin with, the pedagogical aims have been attained quite effectively by Larcher, in our view. The
successive chapters and sections of the book recapitulate the different aspects of the conjugation of Arabic
verbs, most comprehensively and successfully in the case of the functions of the derived measures and
their semantic connection and evolution (chapters III to XI), less so perhaps, we would say, in the case of
voice (chapter II), aspect, auxiliary kna, negatives and other questions dealt with in the final sections
(chapters XII to XVI), which, at least in our opinion, contain more controversial statements.

25. The articles meant are our Marginalia on Arabic diglossia and evidence thereof in the Kitb al-A n, in Journal of
Semitic Studies 20.1 (1975) 38-61, esp. 39, fn. 1, and Littmanns Neues zur altnordarabischen Dialektkunde in Zeitschrift der
deutschen morgenlndischen Gesellschaft 99 (1945-49), 168-180, esp. 168.
26. P. Larcher, Le systme verbal de larabe classique (Didactilangue), Aix-en-Provence 2003, Publications de lUniversit de
Provence, 14,5 x 20,5, 191 pp.



Not surprisingly, the traditional French technical terms accompli, inacompli, passive, active, moyen,
etc., are upheld and defended27 by the author, perhaps as a tribute to the solid grammatical education given
by private and public schools in France, although they are often less appropriate than desirable to reflect
the true oppositions characterizing the Arabic verb. It must be remembered that this particular area of
grammar since its very beginnings was not well served by the native grammatical terminology, modelled
as it was in some points on Syriac and Greek mental patterns, and therefore imbued with concepts at least
partially alien to Arabic grammar, as is obvious in the case of m and muri. Obviously, these two
native coinages do not recommend themselves in a technical treatment of the matter, because neither the
m is a genuine past tense, nor the muri is but partially similar to the nouns in their inflection
(irb), in addition to their lack of semantic symmetry with each other; therefore, the best solution can
only be the adoption of new functionally valid designations.
Our choice, in the trail of other established scholars, has been to borrow perfective and
imperfective from the grammar of Slavic languages, whose verbs share with Arabic the basic opposition
of aspect.28 While it cannot surprise, on logical and statistic grounds, that most perfective actions, i.e.,
conceived as complete processes, do coincide with the past time, and most imperfective actions, i.e.,
conceived as incomplete processes, with the present or future tenses, which helps in understanding why
native Arab grammarians called the perfective m and described muri as the expression of l
(present) and istiqbl (future), it is no less evident that, with that interpretation of the Arabic verb, we
cannot explain the use of the m in optative, jussive, conditional or temporal, correlative and proverbial
utterances, which will be given mere contextual explanations by Larcher in pp. 153 ff., instead of
acknowledging that the aspect simply retakes its upper hand in them, while the tense connotations are
conversely those resulting from the context. Phrases like /braka llhu fk/ God bless you! (p. 154),29 /in
a llh/ God willing (p. 155), /auka i marra lbusr/ I shall come to you when the dates will
be ripe (p. 140), to which we can add the saying /anaza lurru ma waad/ a man of honour always

27. His personal attachment to this terminology is reflected in p. 137, when he says that native grammarians opposed faala
(past) to yafalu (non-past), while the Arabists use accompli and inaccompli, thus stretching to all what in fact is applicable only
to a significant share of the French. He is not happy with the clear-cut and diaphanous perfective and imperfective and
overlooks that accompli and inaccompli cannot be severed from their semantic basic value of accomplishment, i.e., belonging to
the (perfect) past and, therefore, implying a tense system, alien to the original and basic aspectual logemes of the Arabic verb.
28. I.e., Russian vid, which was translated into Western languages, together with the couple sovershennyj perfective and
nesovershennyj imperfective. The quote by Larcher (p. 151) of Qurn VII 44, /wand abu nnri/ and the damned will
shout, etc., given as an example of parfait de prophtie, in front of which even D. Cohen would have recognized the failure of
aspect to account for such a passage, might well be analysed as an exact match of the situation in the Russian verb, where the
future is expressed by the perfective verb, in parallel to the present expressed by the imperfective, both without the mark of past,
which in turn generates the perfect from the perfective stem and the imperfect from the imperfective (e.g., ljublj I love,
poljublj I shall love, ja ljubl I loved, used to love, poljubl I fell in love, loved once); however, it must be acknowledged
that, the aspect systems of Slavic and Semitic not being totally identical, this expression of the future is not characteristic of
Arabic and must not be added to the list of those mentioned below: instead, when the context of that Qurnic quote is duly
checked, we find it at the end of a long chain of perfectives with the connotation of future, triggered in the usual way by a long
temporal clause, beginning in VII 37, /i gathum rusulun ql/ when our messengers will come to them they will
say, followed in apodosis by another chain of similar perfectives (/had/ they will witness, /qlat/ she will say, etc.), i.e.,
the same structure of /auka i marra lbusr/ I shall come to you when the dates will be ripe (p. 140, see below).
29. Needless to say, such optatives are no longer productive in Neo-Arabic and became scarce already in Post-Islamic Arabic,
as we pointed out in Marginalia, p. 53, with an example of substitution of /yaramuka llh/ for /raimaka llh/, in the same way
as /yay lmalik/ sounds nowadays more standard vs. the more Classical /a lmalik/ Long live the king!. But this is not
always the case: thus, for instance, the genuine optative perfective is well and alive in the pan-Arabic /kattar ayrak/ May God
increase your wealth = Thanks.



fulfils his promises and /al amiltu laka t/ Let me make you a crown!30, for the perfective, and
/lim taqtulna anbiya llhi min qablu?/ Why were you killing the prophets in the old time? (p. 141),
/taqlu waqad mla l abu bin ma/ She was saying as the camel saddle tilted under our weight31,
or /qla lah kun fayaknu/ (p. 141) He said to it be, and lo, it was already (existing) for the
imperfective, are all clear examples of the aspect system at work, where none of those perfectives means
past time, but just whole processes, such as wishes and hopes waiting for fulfilment, conditions that must
take full place in order to produce their whole effects, and self-exhortation to action, while the
imperfectives therein do not mean present nor future, but processes in progress, whether durative or
This momentous issue is developed in chapter XII where, in our opinion, Larcher does not hit the
mark, because of his excessive concern over incongruous comparisons with French and other Western
languages,32 as well as of his almost exclusively synchronic approach, certainly dictated to him by the
didactic character of this work, which does not allow him to take into due account the undeniable
evolution from Old Arabic to Modern Standard Arabic from less to more grammatical acknowledgment of
tense nuances,33 leading to a certain erosion of an originally and basically aspect-sensitive verbal system.
It is also obvious, as he says, that tense and aspect do not exclude each other, both of them playing some
role in the Western European languages as well as in Arabic but, in a linguistic analysis, what counts is
which of them is the categorical or logematic axis of the conjugation, of course not excluding the binary
solutions in some languages, and which is only a consequence of lexical, logical or, at any rate, contextual
circumstances.34 It is obvious that, in isolation from any context, /qatala/ always conveys a complete
process, but not necessarily in the past, and /yaqtulu/ always means an incomplete process, but not
necessarily in the past: therefore, the logematic axis of the opposition is aspect, not tense.35
Modal nuances is another complex matter to which Larcher devotes chapters XIII and XIX. While
agreeing with him on most of their contents, especially, the substantial difference between the auxiliaries
of our Western languages, which produce true compound tenses, and the use of /kna/, which he defines
as a verbe oprateur, i.e., roughly a modifying verb, designed to provide a modal nuance, like the

30. From Ibn Idrs Albayn almu rib (ed. G.S. Colin & E. Lvi-Provenal, Leiden, Brill 1951, p. 23).
31. From Imruulqays muallaqah, verse 13.
32. He makes statements such as the Arabic m is similar to the English or the German pasts (p. 139), to the French
pass compos (p. 140), and seems to give equal weight to tense and aspect in the Arabic verb without, however, stating
unequivocally his binary position on this issue.
33. As well as modal ones, which accounts for the great increase of the use of auxiliary verbs in Post-Islamic Arabic, above all
in prose.
34. Thus, for instance, Turkish is not thoroughly devoid of means, lexical, logical or, widely speaking, contextual to express
the feminine gender unequivocally, as demonstrated by the translations from Western languages (e.g., o geldi he/she/it came,
but kz // kadn // kralie geldi the girl // woman // queen came, where Arabic would have marked the gender logeme twice, in
both the subject and the verb, i.e., /at/ vs. /atat/, in the first case, and /atati lbintu // lmaratu // lmalikatu/), but, nevertheless,
gender does not exist at all in Turkish grammar and cannot be posited in it on account of that contextual capacity.
35. This simple and efficient test of relevance has not been applied by the many brilliant scholars who have stuck to the
binary character of the Arabic verb, a very expressive term coined by A. Zaborski in his defence of Kuriowiczs hypothesis,
Kuriowicz and the so-called aspect in Classical and in Modern Arabic, in Analecta Indoeuropea Cracoviensia II, 1995, 529541, to which we cannot concur with our most knowledgeable colleague and good friend. Most Western European languages have
accommodated aspect too in their basically tense systems, while Slavic languages have incorporated tense to their
characteristically aspectual system, but if that test is applied to both cases, it is obvious that a Russian verbal form, whether finite
or non-finite, belongs always to one of the two aspects (= basic feature of the system); by comparison, in a language like French
only some non-compound forms will express aspect, like the imperfect and gerund, while subjective time, i.e., tense (= basic
feature of the system) will always be expressed, with the exception of the infinitive, where neither aspect is present at all.



appropinquation, beginning and uncertainty,36 there are occasional divergences from our views. For
instance, his judgment of total equivalence between /kna yaqmu/ and /kna qiman/ (p. 147): they
may be synonymous in some cases but, otherwise, Fleisch, who had a near-native command of Arabic and
very deep insights in its grammar, was right in understanding il se levait (he used to stand up / was
standing up) in the former example, and il tait dbout (he was standing / up) in the latter. The sens
de langue cannot here be offset by our colleagues grammatical objections, in which we again detect his
insufficient sensitiveness to aspect, since this is attached to the imperfective /yaqmu/, but not to the
participle /qim/,37 in spite of being both alike modified by /kna/, which cannot go without semantic
Our next remark concerns the analysis by Larcher (chapter II) of faila as some kind of middle
voice between active faala and passive fuila. Not only the diathesis active vs. passive does
not exist as such in the Arabic grammar, where there is a mere opposition between verbs with a known
subject to verbs with an unknown subject, as designated by technical terms absolutely clear and functional
in this case (namely,  atu lfil and  atu lmahl), which would make impossible the existence
of a middle voice, but in fact the characteristic implication in the latter of a subject affected by his own
action, while semantically extant in Arabic, is not again a category of its grammar. The traditional label of
stative verbs given by Semitists to both faila and faula, coupled with an eventual differentiation
between temporary and permanent state or condition, and a warning to the effect that many a faila has
evolved semantically and morphologically to the status of the active verbs, exactly like those of the type
faala, was accurate and sufficient and nothing is gained, in or view, from drawing diagrams like those
presented by the author in pp. 23 and 28,38 in which the three vocalization types of agentive perfectives are
combined with that of the non-agentive ones, suggesting an interplay of those types and diathesis which is
not a part of the Arabic grammar.
We have only isolated minor remarks to chapters III to XI, as our suspicion (pp. 35, 55 and 107) that
the entire root {mkn} is a diachronic derivate from /makn/, i.e., a member of the root {kwn}; next, we
have misgivings about a development from the III and VI measures from the II, because insistence would
imply reciprocity (pp. 48-51)39, without excluding occasional phonetic equivalence between /v:2/ and
/v22/, and, finally, we do not believe that /amr/ be an archaic doublet of /nafs/ in the sentence /aslama
amrah lillh/ he placed his affairs in Gods hands. Equally minor also are other issues, especially
diachronic ones, where we hold different opinions, which is only natural and does not imply an error on
either side, but just a different viewpoint. Such is the case of the hypotheses about the development of
causative faala (pp. 35-42), where Larcher demonstrates his familiarity with grammatical theory and
bibliography by offering no less than three kinds of explanations, namely, morphological, syntactic and
paradigmatic, in medieval and modern versions, with their matching arguments and loci probantes. This is
not, of course, the place to discuss each one of them; instead, we feel tempted to add our own hypothesis,
in this case, of a synchronic kind and not devoid of support in the wider frame of Semitic linguistics.
36. Described by W. Wright, A grammar of the Arabic language, II 106-109.
37. As in the case of the non-agentive marb, which we shall mention below, an agentive participle is absolutely devoid of
aspect and/or tense connotations, which explains why, e.g., qri may mean reader, but also literate, capable of reading, and
the motto of the P.L.O., /inn idn/ does not exactly mean we shall return, but we must or can return, we are of the
returning kind, where the emphasis is not on the future time, but on the inexorability of the process.
38. See also p. 58, where the four possibilities are supposed to have been all basic vocalizations of the verb.
39. It is true that Zaborski in his paper Main and secondary functions of derived verbs in Arabic (in Lingua Posnaniensis 48
(2006) 165-189) argues this case very effectively, within a comprehensive explanation of the measures II, III, IX, XI and XV, but
this is done at a much deeper diachronic level than that meant by Larcher. Not only for Arabic, but for the whole South Semitic
branch, the lengthening of the first vowel and the gemination of the second consonant of the stem had become different marks
with different functions, in spite of occasional relics and semantic junctures.



Following the path opened by Rundgren40 in his enlightening study of geminated imperfectives in some
Semitic languages, where he introduced the concept of re-utilization of obsolete morphs in new functions,
we have demonstrated that a large number of the Arabic II measure verbs41 are neither originally nor
semantically true intensives, but have most likely assumed that appearance in the aftermath of the
adoption of North Arabian by former South Arabian speakers, who continued in their creole phase to
geminate the indicative imperfectives without any intensive nuance, although their native North Arabian
neighbours, even their own offspring, may well have introduce it, thus generating instances of reutilization of that morph.
As a matter of fact, the impact of the absorption of a relatively large number of former speakers of
South Arabian, a language with geminated indicative imperfectives, may well have had other
consequences in the North Arabian system of derived measures: it is quite reasonable to assume that a
South Arabian *{yua1a22i3u} (indicative perfective of the causative perfective {a12a3a}, the IV
measure), through the characteristic loss of // in North Arabian, became {yu1a22i3u} and could not be
distinguished from the indicative imperfective of the intensive, II measure). Considering the poor marking
of the IV measure, as a consequence of the weakness and frequent elimination of that prefixed morph, this
could well have been the way through which, II measures began to supplant the IV ones as causatives, in
the same manner as the latter was often superseded by the I,42 since the different vocalization of the prefix,
which is even cancelled in the non-agentive voice, was insufficient to distinguish them morphologically.
Incidentally, the same situation might have led to the confusion of VIII and V measures, conspicuous in
many synonyms as mentioned by Larcher (p. 94), who tries to explain it again through paradigmatic
schemes: here {yat1a22i3u} and {yat(a)1a22a3u} would again, in our guess, have been too similar to
prevent mergers and North Arabian speakers might have developed a free option between {yat1a2i3u} and
Another case of disagreement between us and Lercher is a consequence of his acceptation of a
passive voice in Arabic, in pp. 121-122 where, without using the technical term, he is obviously
analysing /mars/, /mal/, /marm/, /ma fr lah/, etc., as passive participles and, therefore,
past participles, i.e., gard, abandonn and pardonn, which obliges him to explain the idioms
where they appear as cases of transposition nominale de formules, such as /arasah / aalah /
raimahu llhu/ may God protect her / forsake him / have mercy upon him, about which he says that
these idioms are not literally understood as he upon whom one has mercy or who is forgiven, but he
for him one wishes that he be pitied or forgiven. Which is a correct statement, but grammatically
unnecessary, when considering that those participles are not truly passive, in the sense that they would
mean that the grammatical subject is not the actor but the recipient of the action, not past, at all, which is
a category absent from the Arabic verbal system; in fact, the Arabic participles are not only tense-less, but
also aspect-less, so that, e.g., in contemporary Standard Arabic /mun marb/ may mean drunk water,
but also, as is common, drinking water, i.e., water than can eventually be drunk, and a cellular phone
is /maml/, not because it is carried about at all times, but because it may be so, i.e., it is portable.
Some other areas of lesser disagreements or comments are found in p. 124, where the author implies
that the XI measure exists only in Moroccan, within modern Arabic dialects, while, in fact, is
40. In his trend-setting essay Intensiv und Aspekt-Korrelation. Studien zur thiopischen und akkadischen Verbalstammbildung, Uppsala-Wiesbaden, Lundequistka, 1959.
41. See our paper Geminate imperfectives in Arabic masked as intensive stems of the verb, in Estudios de Dialectologa
Norteafricana y Andalus 8 (2004) 33-57 (Homenaje a Peter Behnstedt en su 60 aniversario).
42. See our From Old Arabic to Classical Arabic through the pre-Islamic koine: some notes on the native grammarians
sources, attitudes and goals, in Journal of Semitic Studies 21.1-2 (1976) 62-97, esp. 86 and fn. 1, where we mention Sbawayhi
and Alalls survey of the confusion of I and IV, as well as II and IV measures.



characteristic of the entire Western branch,43 and in p. 132, where lam cannot be a direct derivate of
lm, but of Aramaic lm.
Finally, in chapter XVI, we must take exception to the assumption of functional differences between
the negations /m faala/ and /lam yafal/. Larcher is more correct, in or view, when he says (p. 162)
that the latter is a case of survival of the Old Semitic accompli, i.e., the prefix conjugation, not yet
necessarily governed by the aspectual logeme, as well as when he establishes the syntactic contexts in
which both structures are found most often or correctly. However, the diachronic character of this issue
eludes him again, as he does not tackle the etymological question, namely, that the negative /m/ is always
the stylistic evolution of an underlying rhetorical question (e.g., /m faal?/ what [do you say] he did?
> he did not), while /lam yafal/ indeed contains a fossilized survival of the Old Semitic suffix
conjugation, preceded by a contraction of /la-m/ certainly not.44 Larcher is again right in presuming
that this negative particle developed in conditional structures (p. 161-162), although he appears to miss the
central point, namely, that the presence of the assertive /la-/ is characteristic of such structures, as a
consequence of the perfective mutual conditioning and concatenation between requirement and outcome
(if A happens, then certainly B follows). Otherwise, however, once both negations existed side by side,
they came to be used as synonymous, merely subjected perhaps to dialectal or idiolectal preferences.
A second-rate issue to which we acknowledge our being particularly sensitive, is the transcription
system labelled by us as the Arabists Arabic, i.e., the habit of using pausal forms everywhere, except in
the cases where final vowels are obvious, e.g., perfective verbs and functionals, thus creating an artificial
Arabic, neither Classical nor dialectal, nor anything having ever existed. Larchers transcriptions are
correct most often, for instance, when he quotes from old poetry or from the Qurn, but not always, as in
his excerpts from prose, above all, modern, as can be seen in pp. 25, in a grammatical text, 40, where
*afat al- ru, (with irb) is given as Old Arabic, instead of /afati l- ru/, and p. 50 in a
quotation from a newspaper. One of the pitfalls of that system, is the uneasiness with certain kinds of
words, as can be seen in p. 139, where the author transcribes /fatt amla/ and translates jeune fille,
although in the spoken Arabic of our time this would sound /fat amla/, while the former expression
would be understood as amlas girl.
Printing or copying mistakes are reasonably few; we have detected the following: pp. 35 *makkanuhu
for /makkanah/, 36 * allabna for / allbna/, *maari for /maarin/ (at the end of the hemistich,
not of the verse!) and /biawfari/, better than /aami/, after the received text, 66, where the place
name al-is becomes le Hassa in its French rendering, 93 *adaha for /adala/, 100 *inqaaa
for /inqaaati lkahrab/, 110 *aartu for /artu/, 111 *tanabb for /tanabbaa/, 120 *Ri
and *Rish for the feminine personal name /raa/, and 132 IIIe forme trilitre for quadrilitre.
We hope that this rather long list of remarks will not be interpreted as a negative judgment on
Larchers work. It is only natural and positive that scholars disagree on some points, which in turn may
spur new efforts to clarify the issues and bring about better solutions to our problems, even mere better
wording in their texts, so that they become clearer. A long review is always a witness to the importance
attributed to a book and a contribution, or at least the wish to make it, to the authors opinions.

43. Cf. H.R. Singer, Grammatik des arabischen von Tunis, 1984, 393-397, E.F: Suttcliffe, A grammar of the Maltese
language, 1936, 98, and our A grammatical sketch of the Spanish-Arabic dialect bundle, 1977, 105, fn. 166.
44. Cf. the parallel /kam/ how much < */ka-m/, and the allomorphs /lim/ < /li-m/ and /bim/ < /bi-m/ (see Wright, A
grammar of the Arabic Grammar I 274).